HarvestChoice generates knowledge products to help guide strategic investments to improve the well-being of poor people in sub-Saharan Africa and South Asia through more productive and profitable farming. To do this, a novel and spatially explicit evaluation framework is being developed and deployed.
Currently the main source of daily historical weather data used in HarvestChoice is the NASA Prediction of World Energy Resources (POWER) Agroclimate Database. The POWER database provides satellite remote sensing-estimated daily weather variables (e.g., minimum and maximum temperature, rainfall, and solar radiation) globally at 1-degree grids, including:
- Solar radiation (daily total): From July 1983 through June 2005; and July, 2006 through current with one month delay
- Air temperature (daily average, minimum, maximum): From January 1983 through December 2006
- Dew point temperature (daily average): From January 1983 through December 200
HarvestChoice also uses a longer-term climate database for cropping systems characterization as well as assessing risks associated with climate variability. The main source of these long-term climate data is the University of East Anglia Climate Research Unit Time-Series (CRU-TS) Global Climate Data 3.0: Reformatted for Spatial Analysis by CGIAR-CSI. The CRU-TS dataset has monthly time-series of climate variables, for the period 1901-2006, covering the global land surface at 0.5 degrees resolution. Rainfall (daily average): From January 1997 - current with a 2-month delay.
While HarvestChoice is a secondary user of weather and climate data, the project is working with partners to downscale existing data sources to generate weather/climate variable estimates at a higher resolution (http://labs.harvestchoice.org/2010/08/slate/), e.g.,
- Temporal downscaling of (1901-2006) CRU 30 arc minutes (0.5 degree) time-series climate data from monthly to daily resolution, and
- Spatial downscaling of (1997-2008) NASA POWER daily weather data from 60 arc minutes (1 degree) to 5 arc minutes (10 km) resolution, using elevation as a covariable.
- Converging both weather databases (i.e., CRU and NASA) into one single, daily, high-spatial resolution database (Synthesized Long-term Weather [SLATE]; previously known as CRU-Mashup), formatted to be used as input to the crop simulation models.
Four different crop models are currently being used in HarvestChoice. Some models perform better than others in specific contexts, e.g. when applied to particular climates, crops, cropping patterns/rotations, soil quality indicators, and potential management interventions. DSSAT, APSIM , ORYZA, WOFOST. Results of crop modelling will be evaluated against high-resolution sub-national agricultural statistics data, and databases of region-wide field trials managed by agricultural research institutes, including CGIAR centers and FAO. HarvestChoice georeferences and compiles the dataset in an easily accessible format to be used in a range of crop systems modeling platforms (e.g, ICASA format for DSSAT model). For instance, a map has been developed showing the location of CIMMYT maize trials in sub-Saharan Africa. Each trial location reports the performance of local and improved varieties in response to a range of treatments, including their tolerance to local biotic/abiotic constraints (http://mapspam.info/).